Skip to content

AI in Healthcare and Emerging Technologies (Part I): Digital Twin and Digital Thread

AI in Healthcare and Emerging Technologies (Part I): Digital Twin and Digital Thread

“She is your mirror shining back at you with a world of possibilities. She is your witness who sees you at your worst and best but loves you anyway. She is your partner in crime, your midnight companion, someone who knows when you are smiling, even in the dark. She is your twin sister.” 

Unknown 

This newsletter series will focus on emerging or existing technologies that will leverage artificial intelligence to maximize the dividend in healthcare. We will start with digital twin and digital thread:

Digital twin (first coined by NASA’s John Vickers in 2010) is the concept of a dynamic virtual representation of an object or objects (such as a pacemaker or equipment in a room) or a system (such as an operating room or a device manufacturing process) used for analysis, usually to improve on the real-world version and/or to minimize unfavorable outcomes.

Perhaps one of the first experiences with such a concept was in 1970 with the Apollo 13 mission during which NASA engineers duplicated the spacecraft in trouble to determine how to best find solutions to the onboard problems. In addition, digital twin is different than simulation in that the former is actually a virtual environment with real-time data. A digital twin, in short, is a dynamic digital database. Digital thread, a related concept, creates a close interface between digital and physical worlds in order to optimize processes and is a connecting element of a digital twin.

The very exciting new development in this area has been the potential use of digital representations in healthcare:

First, a digital representation of a hospital or a department can be set up for stakeholders to analyze operational efficiencies and capacities with different allocations of resources. This virtual model can also be used for changes in condition such as a pandemic or other healthcare emergencies. Second, a digital twin can also be applied to medical equipment and devices such as a ventilator or a cardiopulmonary bypass circuit to predict potential malfunction in the future.

In addition, similar to the marooned Apollo 13, a patient with a disease or condition can be analyzed with digital twins that will be available for clinical trials without other humans. A patient with heart failure, for example, can have his/her heart data duplicated for a myriad of interventions such as biventricular pacing for heart failure or even various medications with pharmacogenomic information prior to actual institutions of these therapeutic interventions.

This strategy is essentially a realistic virtual model of a human organ or body part in order to analyze this organ or body part in various conditions coupled to therapeutic interventions to maximize the survival of the organ or benefit of the body part.

These digital twin simulations are difficult to deploy as they require domain expertise and realistic feedback mechanisms as well as real-time, real-world data. These simulations will become increasingly more complex in biomedicine with the advent of medical wearable devices and sensor technology that will generate a tsunami of biomedical data.

Machine learning, deep reinforcement learning (the AI tool used in AlphaGo by DeepMind) in particular, can help to make more accurate predictions in digital twin patient simulations.

Show Buttons
Hide Buttons